Full Citation

Title: Differentially Private Data Publishing

Citation Type: Dissertation/Thesis

Publication Year: 2021

ISBN:

ISSN:

DOI:

NSFID:

PMCID:

PMID:

Abstract: The rapid development of information technology has opened up the era of big data. A large number of high-dimensional and heterogeneous data appear in practical applications, which are often published to third parties for data analysis, recommendations, targeted advertising, and reliable predictions. However, publishing these data may disclose personal sensitive information, resulting in an increasing concern on privacy violations. Privacy preserving data publishing (PPDP) has gained significant attention in recent years as a promising approach for information sharing while preserving data privacy [1]. Differential privacy has gradually become the de facto standard privacy definition and provides a strong privacy guarantee. In this dissertation, in my efforts to fill the gap between the differential privacy theory and its applications, we target to design multiple differentially private data publishing mechanisms addressing the challenges for high-dimensional, high-volume and high-variety data. To be specific, we study this problem from the following three aspects. First, we provide intuitive interpretations and illustrations on the important ideas in differential privacy, especially noise calibration to global sensitivity and smooth sensitivity, and composition properties. Taking social networking as an example to study how to adapt. differential privacy from tabular data to social network data, we explore the interplay between differential privacy and social network analysis by systematically introducing four models of differential privacy definitions, then we review existing differentially private methods for three most widely-used graph analysis techniques, and put forward a research agenda that involves four open challenges in differentially private data publishing. Second, differentially private publishing of high dimensional data remains a challenging problem – it suffers from the “Curse of High-Dimensionality” [2]. Most existing approaches generally ignore the different roles a dimension may play for a specific query-one dimension may be more important than another for a particular query. Additionally, one dimension may release more information than another if the same amount of noise is added; thus evenly allocating the total privacy budget to each dimension degrades the performance. In order to address these challenges, we propose a differentially private high-dimensional data publication mechanism (DP2-Pub) that runs in two phases of a Markov-blanket-based cluster learning and an invariant post randomization (PRAM). We also extend our DP2-Pub mechanism to the distributed scenario with an untrustworthy central server. Third, conventional private data publication mechanisms aim to retain as much data utility as possible while ensuring sufficient privacy protection on sensitive data. Such data publication schemes implicitly that all data analysts and users have the same data access privileges levels. However, it is not applicable for the scenario that data users often have different levels of access to the same data, or different requirements of data utility. The multi-level privacy requirements for different authorization levels pose new challenges for private data publication. Traditional PPDP mechanisms only publish one perturbed and private data copy satisfying some privacy guarantee to provide relatively accurate analysis results. To find a good tradeoff between privacy preservation level and data utility itself is a hard problem, let alone achieving multi-level data utility on this basis. In this work, we address this challenge in proposing a novel framework of data publication with compressive sensing supporting multi-level utility-privacy tradeoffs, which provides differential privacy.

Url: https://www.proquest.com/docview/2509629932/fulltextPDF/D5E3312006BE4DE3PQ/1?accountid=14586

User Submitted?: No

Authors: Jiang, Honglu

Institution: The George Washington University

Department: Philosophy

Advisor:

Degree:

Publisher Location: Washington, D.C.

Pages: 1-97

Data Collections: IPUMS USA

Topics: Population Data Science

Countries: